1. Field
The present invention relates to systems for determining consumer preferences. More specifically, the invention relates to self-explicated trade-off analysis systems used to quantify preferences of a respondent with respect to product attributes and to product attribute levels.
2. Discussion
Manufacturers are presented with many choices during the design of a product. For example, a manufacturer must choose from among several available product features, or attributes, when deciding which attributes to include in a product. Some attributes are optional while others must be included. In the case of a television set, “Chassis color” is an attribute that must be included and “Picture-in-picture” is an optional attribute. For each included attribute, a manufacturer must also choose an attribute level to associate with the attribute. Examples of attribute levels which may be associated with the attribute “Chassis color” include “black”, “white”, “blue”, etc.
Occasionally, a manufacturer produces several versions of a similar product by varying product attributes and/or attribute levels among the several versions. In such a case, the manufacturer must determine attributes and associated attribute levels to include in each version as described above. Moreover, the manufacturer must determine how many units of each version will be produced. For example, a manufacturer choosing to produce televisions having a black chassis and televisions having a blue chassis must also determine how many of each type of television to produce and offer for sale.
Product pricing represents a further choice for product manufacturers. In this regard, a manufacturer attempts to choose a price for each produced product that will maximize overall profit to the manufacturer. Of course, price may also be considered a product attribute, with associated attribute levels consisting of particular prices.
Each of the foregoing choices, as well as other choices, may be greatly facilitated if the manufacturer has detailed and accurate information relating to consumer preferences. A consumer, in this regard, is any entity to which a product (i.e. a good and/or service) may be offered. Such consumers include individuals, businesses, and purchasing managers. Consumer preference information can be used to determine the popularity and desirability of particular product attributes and attribute levels to consumers. Therefore, by using this information, a manufacturer is more likely to choose product configurations as well as production amounts and prices for each product configuration that maximize overall profit.
In view of its importance, manufacturers expend significant resources in their attempts to obtain detailed and accurate consumer preference information and to analyze marketplace choices. These resources are most commonly allotted to conventional consumer surveys. Such surveys typically consist of a list of predetermined questions designed to elicit information from a respondent regarding the respondent's feelings toward products, product attributes, and product attribute levels. Surveys may be administered randomly, for example by stopping respondents at shopping malls or other retail areas, or by contacting specific respondents who are targeted because they are members of a demographic group about which information is desired.
Conventional surveys present several inherent drawbacks. First, since survey results are compiled into general demographic categories, surveys merely determine, at best, preferences of a theoretical average consumer belonging to each demographic category. Accordingly, survey results are only marginally correlated to any one respondent's preferences. Therefore, such results lack predictive precision of a particular respondent's preferences with respect to marketplace choices that are available and not yet available. Second, although conventional surveys may indicate whether one attribute level (e.g. “black chassis color”) is generally preferred over another attribute level of the same attribute (“white chassis color”), such surveys do not provide any reliable means for comparing preferences across attributes. For example, conventional surveys are generally unable to determine the degree to which a respondent prefers a black chassis to another color so as to enable comparison between that degree and the degree to which the respondent prefers a 27″ screen to another screen size. As a result of these drawbacks, conventional surveys are poor at producing useful quantified preference information of individual respondents.
Focus groups are another conventional vehicle used to obtain preference information. In a typical focus group, certain respondents are randomly selected (or selected based on demographics as described above) to answer questions and/or to participate in a group discussion regarding a product or a type of product. Answers and comments put forth by the respondents are noted and tabulated to create preference information similar to that obtained using survey techniques. However, because of their interactive nature, focus groups may elicit information which is more pertinent than that elicited by surveys. Despite this possible advantage, focus groups still suffer from the drawbacks described above with respect to conventional surveys.
The field of trade-off analysis was developed to address the above and other shortcomings in conventional techniques for determining preference information. Generally, trade-off analysis techniques attempt to quantify a respondent's preference for a particular product's attributes and attribute levels. Such quantification is intended to allow a manufacturer to easily and accurately compare the attractiveness of various product configurations to a respondent. For example, many trade-off analysis techniques allow a manufacturer to compare the attractiveness of a 27″ television with Picture-in-picture capability priced at $399 with that of a 35″ television with a digital comb filter priced at $599. This comparison is possible because the techniques associate a particular numerical value with a respondent's preference for each attribute and attribute level. Accordingly, the relative attractiveness of any attribute or attribute level with respect to any other attribute or attribute level can often be determined simply by comparing the appropriate associated numerical values.
According to one classification scheme, four types of trade-off analysis techniques exist: conjoint; discrete choice; self-explicated; and hybrid. Conjoint analysis generally requires a respondent to rate or rank product configurations with respect to one another. Typically, the consumer is asked to evaluate twenty to thirty product configurations. Each configuration includes different combinations of attributes and attribute levels being evaluated. By appropriately varying the configurations, a regression model can be estimated for each respondent in order to estimate respondent-specific numerical values for the attribute levels.
Conjoint analysis is an improvement over conventional systems for determining preference information. For example, determining preferences by observing respondent behavior is difficult or impossible because respondent behavior can usually be observed only with respect to a few combinations of attributes and attribute levels (i.e., the combinations that exist in the marketplace). Accordingly, it becomes difficult to separate and distinguish between the preferences of different consumers and to predict effects of changes in attributes and/or attribute levels on respondent behavior. On the other hand, conjoint analysis allows for improved learning of respondent preferences through controlled variation and controlled co-variation of attributes and attribute levels.
According to discrete choice analysis, a respondent is presented with a set of product configurations and asked to select either the configuration that the respondent is most interested in purchasing or no configuration if the respondent is not interested in purchasing any of the presented configurations. The process is then repeated for other sets of product configurations. In contrast to conjoint analysis, which may be used to estimate a regression model for individual respondents, discrete choice analysis may be used to estimate a mixture method (similar to a regression model) for a group of respondents.
While conjoint analysis and discrete choice analysis determine respondents' preferences for particular attribute levels of associated attributes indirectly, self-explicated analysis directly determines preferences by asking respondents how important each product attribute range and attribute level range is to their purchasing decisions. According to some self-explicated analysis models, respondents are presented with all attributes and attribute levels to be evaluated, and asked to identify attribute levels that are unacceptable. An unacceptable attribute level is one that, if included in a product, would cause the product to be completely unacceptable to the respondent, regardless of any other attributes and attribute levels Included in the product. For example, a respondent may indicate that an automobile including an attribute level of “pink” associated with the attribute “color” is completely unacceptable for purchase regardless of any other attributes or attribute levels included in the automobile. Accordingly, “pink” is identified as an unacceptable attribute level for that respondent.
Next, the respondent is asked to identify, from the acceptable attribute levels, the most-desirable and the least-desirable attribute levels associated with each presented attribute. For example, assuming that the respondent's most important attribute has a rating of 100, the consumer is then asked to rate the relative importance of each remaining attribute from 0 to 100. Of course, scales other than 0 to 100 may be used. Next, for each attribute, the desirability of each attribute level is rated with respect to all other acceptable attribute levels of the attribute. The respondent's preference for an attribute level is then obtained by multiplying the relative importance of its associated attribute by its desirability rating.
Hybrid analysis techniques utilize a combination of features from the above-described techniques. The most common example of a hybrid analysis technique is Adaptive Conjoint Analysis (ACA), a product of Sawtooth Software, Inc. According to ACA, a respondent is taken through rankings of attribute levels and ratings of relative attribute importance (similar to self-explicated techniques) and then asked to identify, for each of a series of pairs of product configurations, which one of the pair is the most desirable and the degree to which it is more desirable. Other examples of hybrid models include the Cake Method and the Logit-Cake Method developed by MACRO Consulting, Inc.
Each of these trade-off analysis techniques requires respondents to provide consistent, thoughtful responses to presented inquiries. A respondent may be able to provide such responses if presented with a small number of inquiries, but is unlikely to do so if presented with many inquiries. In this regard, the number of inquiries presented by each of the above techniques increases sharply as the number of evaluated attributes and/or attribute levels increases. Such an increase in the number of inquiries also causes a corresponding increase in the amount of time required to answer the inquiries. Therefore, as more attributes and attribute levels are evaluated, various forms of respondent bias are likely to increase, such as confusion, a waning attention span, a lack of time, a lack of patience, boredom, and haste. These increased respondent biases result in increased respondent error and inaccurate preference information. Also increased is a respondents tendency to abandon the process and to simply cease answering further inquiries, in which case the resulting preference information is partially or totally unusable.
Another form of respondent bias is caused by consumer attitudes toward particular attributes and/or attribute levels. As described above, conventional trade-off analysis techniques ask a respondent to evaluate the importance of an attribute or attribute level with respect to other attributes or attribute levels. However, if the respondent has an extreme dislike for one of the attributes or attribute levels, the consumer may overestimate the importance of the other attributes or attribute levels. Moreover, a relative importance of an attribute with respect to the difference between its best and worst attribute levels may differ based on a number of intermediate attribute levels presented to a respondent.
In view of the foregoing, what is needed is an efficient trade-off analysis system to quantify consumer preference information. Such a system may or may not address the forms of consumer bias experienced by conventional systems, produce accurate and useful preference information with respect to product attributes and attribute levels, and/or improve manufacturer and/or retailer choices to maximize profit for the manufacturer and/or retailer.